...
\uD83D\uDC65 Participants
Lancaster: Gerard, Steven, Matt
Glasgow: Sam
- Boards
| Jira Legacy | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|
|
...
\uD83D\uDDE3 Discussion topics
Item | Topic | Presenter | Notes |
|---|---|---|---|
XRD-6 | Lockless reads |
| |
XRD-11 | Slow deletions | LHCb existing issue with slow deletions (up to 100s) of seconds. Ian to run dedicated tests within the RAL controlled environments. Test runs using python multi-processing. Tested so far with 5GB files. | |
N/A | Corrupted LHCb file(s) | Caused by write conflict between davs and gridftp. A variety of problematic outcomes observed from replication test. Xrootd/davs takes a couple mins to clean up after reporting failure | |
Proposal: On unlink call, remove any existing checksum first? | Discussion suggested this might be a more complex situation and warrants further though. Not mixing protocols might mitigate these situations (for the most part … ) | ||
|
|
Site reports
Glasgow
NTR (other than that I'm planning to push all our Xrootd services to 5.4.3-release when it finally gets an rpm)
It might be worth noting that I think Brunel are fairly advanced on their Cephfs/xrootd testing with CMS [although Raul doesn't come to *this* meeting]
Lancaster
Ran over our redirector plan in this morning’s Storage Meeting. Plans on hold until after Matt is back from the training week. Current simple setup appears to be holding up quite well. Will also upgrade to 5.4.3.
Several of us at Lancaster are now “equipped” to assist with any rucio testing.
Less interesting cephfs continues to replace all the dodgy NFS uses at the site, swapped out the arc session directories.
✅ Action items
⤴ Decisions
...